Backgrounds

We are a team of active students in the mobility program. My team and I have had several new experiences, starting with small things in campus life, one of which is shopping at the canteen. My colleagues come from different campuses, starting from Campus A, Campus B, and myself from Campus C.

In this project, I will attempt to analyze how active students in the student mobility program perceive the service, cleanliness, and food quality at the University cafeteria, and analyze the factors influencing satisfaction based on questionnaire data.

Datasets

This data was obtained from an internal survey on student satisfaction levels regarding service, cleanliness, and food products at the Binus Anggrek Campus cafeteria. The survey was conducted anonymously using a Likert scale in bahasa language.

DISCLAIMER
The data has been anonymized by removing personal identity information such as names, email addresses, and student ID numbers in accordance with personal data protection principles. The data will be extracted to a .csv file format in every type of changes in the process

Data Cleaning & Preprocessing

Identifies Missing Value and The Causes

## Missing values is founded in:
## major : 6 missing value(s)
## home_campus : 4 missing value(s)

The main reason for the missing values, especially for Home Campus, is that the respondents are not active participants in the mobility program. Since the purpose is to measure the satisfaction of students who are actively involved in the mobility program, I decided to remove these entries with missing Home Campus values.

For column Jurusan, I imputed the missing values using the most frequently occurring major. This approach is intended to preserve useful responses, as the remaining data may still contribute valuable insights for the analysis.

## Missing values after deletion and imputation:
## major: 0
## home_campus: 0

From the results above it shows a successful attemp to handle missing values

Handle Inconsistent Values in Likert Scale

To anticipate inconsistent values, I’ll change all the likert categories to a lowercase (in case there are some miss type in the questionnaire form)

## Before mutation: Setuju, Sangat Setuju, Tidak merasakan, Sangat Tidak Setuju
## After mutation: setuju, sangat setuju, tidak merasakan, sangat tidak setuju

This process is to anticipate errors in encoding process, since the inconsistent value is completely handled, encoding is needed to transform the likert category to a numeric form of scale.

Separating Date, Timestamp, and Durations

By default, the tool used to gather questionnaire results has a time stamp about starting time to a completion time. Where in this case I will parse them to an identical column. In this case I’ll separate the date and time into an individual column.

Encoding Likert Scale and frequency

## Unique value for every likert scale:  setuju, sangat setuju, sangat tidak setuju, tidak setuju, tidak merasakan

The result above are unique values. Where in this case, every likert scales will be encoded manually to a range number from one(1) to five(5).

## Unique value for every likert scale (after encoding):  4, 5, 1, 2, 3

New Column for Further Analysis

Average/Mean Satisfaction Score
Here I make a new column to support further analysis and prediction tasks. This value is calculated as the average of every encoded likert scale. Why is this important? it provides a single quantitative measure that summarizes overall student satisfaction. It allows for easier comparison between respondents and can be used as a target variable in predictive modeling or segmentation analysis.

Average/Mean Score in Every aspects
To perform an in-depth analysis of the performance of each metric measured such as services, cleanliness, and product, I created a new column to store the average of each question to measure respondents’ satisfaction with the aspects

Analysis

Distribution of Duration

The first analysis is to see the distribution of duration. This graph might help to evaluate respondents who fill out the questionnaire too quickly or too slowly, which indicates their seriousness in answering the questionnaire.

Based on the histogram above, respondents are more likely to complete the questionnaire in about 2 minutes. This data shows that respondents answered the questionnaire fairly quickly.

At 0 minutes, this could be caused by some respondents answered the questionnaire very quickly and might not read the questions (not taking it seriously). Further analysis will be explained through the descriptive statistics below.

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   0.520   2.277   3.065   5.427   4.082  86.400

The shortest duration for respondent to fill questionnaire is 0.520 minutes.

Canteen Visitors Frequency

In this analysis, I’ll try to measure visitation rate with satisfaction. In order to categorize the satisfaction level, I’ll use discretization or binning to make a category out of mean satisfaction score. The mapping of this process are as follows:

(MSS = Mean Satisfaction Score)

  • MSS less or equal to 1 = “Bad”
  • MSS greater than 1 AND MSS less or equal to 3 = “Quite Bad”
  • MSS greater than 3 AND MSS less than 5 = “Good”
  • MSS equal to 5 = “Excellent”

Graph View

After discretization or binning process, the analysis will be utilizing a plot to show the Amount of Canteen Visitors by Frequent Visiting and Satisfaction Category

Explanation:

  • The visitors mostly come once until two times a week, meaning that the canteen is not a frequent place to visit. The overall satisfaction level is on “good” condition with 2 respond show an excellent satisfaction level.
  • Respondent who chose “0 kali” or rarely go to canteen showing the most “quite bad” satisfaction level among others. This might indicated by some people that had a unfulfilled expectations.

Table View

Performances in every aspects

This analysis is conducted to know the performance of every aspects that measured. The aspects that are measured such as, service, cleanliness, and product. The graph below shows the total average of the average performance scores for each aspect. This graph is expected to provide an overview of aspects that can be improved in the Binus Anggrek cafeteria.

Explanation

  • Based on the graph above, respondents gave fairly good responses to the service, as indicated by the average total score falling into the “good” category.
  • This was followed by the performance of the canteen food, which received a “medium” rating, with room for improvement in terms of product quality and price.
  • Finally, the aspect of cleanliness requires significant improvement in terms of cleanliness and access to washing facilities.

Seeing the Weaknesses of each Aspect

In this analysis, I will focus on aspects that have a “Need Improvements” status which referring to Product and Cleanliness aspect. Here, I will create a descriptive statistical table to dig deeper into the areas that are weaknesses that greatly affect the average student satisfaction score for the Binus Anggrek cafeteria.

In this process, it’ll require binning technique to categorize each variables based on their average scores. The Binning process will be mapped as follows:

  • mean less than 3.5 = “Needs Improvement”
  • mean less than 4 = “Medium”
  • mean greater than or equal 4 = “Good”

Graph View

Cleanliness

The average of cleaning_facility variable have the lowest results such as 3,62 (if rounded by 2), followed by neat_furniture_placement with result of the average is 3,69. These both variable are the weak point for cleanliness aspect. This weaknesses also affecting the total average score of cleanliness (could be found at “Analysis” tab) to be the lowest among all existing aspects.

Both of those variable have a high standard deviation (sd > 0,5) which indicate that the distribution was quite wide among the respondents’ answers. Moreover, there was a change in the location of the cafeteria during the collection of the questionnaire results, it might affect the wide response that has mentioned before.

Product

In Product aspect, there are two different variables that fell into “Medium” category, such as affordable_price with the average of 3,65 and price_vs_quality with an average of 3,88. The affordable_price variable had a high standard deviation (~0,93), this indicates a fairly wide spread among respondents’ answers and may be influenced by various segments available.

Services

Finally, most respondents stated that service aspects fell into the “Good” category, as seen from the average of each variable, which was greater than 4 (“Agree” on the Likert scale) and the consistency of respondents (median) answering 4 on most service aspects. However, despite the ratings falling into the “Good” category, staff_response falls into the “Medium” category, as not all respondents felt that the staff in the cafeteria were responsive to questions or complaints.

Table View

Segmentation Based on Home Campus

This analysis was conducted to gain insight into the responses of respondents from various campuses regarding three aspects that influence student satisfaction with the Binus Anggrek cafeteria.

Explanations:
Based on home campus segmentation, the majority of students from Binus Bandung and Binus Semarang have a satisfaction category classified as “Good,” with a few of them falling into the “Excellent” category. On the other hand, some respondents from Binus Semarang fall into the “Quite Bad” review category, which may be due to their lowered expectations regarding the Binus Anggrek canteen. Based on the above results, although satisfaction levels are relatively stable, there is room for improvement to make the experience truly exceptional.

Average Score per Aspect Grouped by Home Campus

This analysis was conducted to determine the level of satisfaction of respondents from different home campuses with regard to the three aspects of the cafeteria.

Graph View

Main Insight:

  1. The Service aspect (services_score) consistently received the highest score across all campuses, with the highest score belonging to Binus Malang (4.112) and the lowest to Binus Greater Jakarta (4.000).
  2. Cleanliness (cleanliness_score) shows the greatest difference between campuses:
    2.1. Highest score: Binus Malang (4.008)
    2.2. Lowest score: Binus Semarang (3.754), this score also caused this campus to be the only one in the “Needs Improvement” category in the previous analysis.
  3. Products (product_score) tend to be in the middle, relatively consistent with small differences between campuses (~0.1).

Explanation for Binus Greater Jakarta: The graph above shows a comparison of scores for each aspect of the Binus Anggrek cafeteria assessment, grouped by home campuses. Some of the data above cannot be confirmed for accuracy, particularly for respondents from the Binus Greater Jakarta campus, as there was only one respondent who scored a 4 on all three aspects. Therefore, the results of respondent from greater jakarta should not be used as a basis for decision-making.

Table View

Segmentation Based on Major

This analysis was conducted to gain insight into the average satisfaction scores of students grouped by major. Therefore I’ll check the unique values in case there were so many majors are in the datasets.

##  [1] "digital business"                       
##  [2] "computer science"                       
##  [3] "public relations"                       
##  [4] "communication - marketing communication"
##  [5] "information systems"                    
##  [6] "data science"                           
##  [7] "computer science - software engineering"
##  [8] "artificial intelligence"                
##  [9] "cyber security"                         
## [10] "game application and technology"        
## [11] "interior design"                        
## [12] "computer engineering"

There are 12 unique values in major that might require to turn them into a broad categories. For example:

  • Tech / IT: Computer Science, Software Engineering, AI, Cyber Security, Computer Engineering, Data Science, Information Systems
  • Design / Creative: Interior Design, Game Application
  • Business / Comm: Digital Business, Marketing Comm, Public Relations

Graph View

Insights & Explanations:
Based on the segmentation of grouped majors, students from the Tech field dominate the respondent population (n = 95), making the scores from this group the most representative reference.

On the other hand, the Design and Business/Communication groups showed relatively higher satisfaction scores. However, due to the very small number of respondents (n = 2 and n = 5), these results are not sufficiently robust to be generalized. This finding underscores the importance of balanced representation in surveys to ensure that the analysis reflects the entire student population.

Table View

Conclusions 

The analysis shows that student satisfaction levels vary widely in every aspect. The previous analysis stated that most respondents answered the questionnaire within 2 minutes, which indirectly shows the seriousness of the respondents in filling out the questionnaire. From the analysis of the cafeteria assessment based on each aspect, there are two aspects that must be given special attention, namely cleanliness and product. Based on the analysis results, there is a significant spread in the aspects of cleanliness and product, where students have diverse answers based on their home campus and major category.

Summary of Key Findings

Some key points are achieved from previous analysis that are described as follows:

  • Majority of respondents are dominated by student in tech field, showing analysis results based on the perspective of tech students.
  • Cleanliness aspect had the lowest average score compared with the others. This showed us that cleanliness must be prioritized with such development.
  • The variation range of responses are quite big, showed by some of the variable (e.g., cleaning_facility) had more than 0.5 (~0.96) for standard deviation. This variation is affected due to canteen that changes location during the survey period. 
  • Home campus segmentation show a significant differences between average score. This clearly showed that student from Binus Semarang dominates the questionnaire respond with a bit high on variability (sd = ~0.57 until ~0.67).

Prioritized Aspect for Improvement

In these three aspects (Service, Cleanliness, Product), there are some variables that had a space for improvement which will be explained as follows:

Cleanliness

Cleanliness aspect have the lowest average score by ~3.88 and fell onto “Need Improvement” category. This aspect has several variables related to it score. The following variables fell into the “Medium” category with plenty of room for improvement in terms of canteen performance in the cleanliness aspect:

Prioritized Variable for Cleanliness

  • cleaning_facility (“The cafeteria provides facilities such as sinks, tissues, and adequate hand washing facilities.”)
  • neat_furniture_placement (“The layout and arrangement of tables/chairs in the cafeteria ensure orderly queues.”)
  • cozy_seating_area (“The cafeteria has a comfortable and adequate seating area.”)
  • cozy_room_layout (“The layout and arrangement of chairs and dining tables create a comfortable atmosphere in the cafeteria.”)
  • clean_area (“The cafeteria area is clean and well organized.”)
  • clean_utensils (“The cleanliness of eating utensils (plates, glasses, spoons, etc.) is maintained.”)

Product

Despite cleanliness aspect who had the lowest avg.score, there is product aspect that has slight higher avg. score and fell onto “Medium” category. As explained before, each aspect had its own variables that fell into “Medium” category with a room for improvement of product aspect.

Prioritized Variable for Product

  • price_vs_quality (“The taste of the food/drinks in the cafeteria is as expected.”)
  • affordable_price (“The prices of food and drinks in the cafeteria are affordable for students.”)

Services

At last, service aspect who had the highest avg. score. Despite being the highest, this aspect also need a bit of change in one variable in order to upgrade the quality of cafeteria services. The mentioned variable is explained below:

Prioritized Variable for Services

  • staff_response (“Canteen staff respond promptly when receiving orders or complaints.”)

Main Recommendation

The cafeteria has weaknesses in terms of cleanliness and the product it offers. In order to confront this challenges, some changes need to be prioritized in both aspects who had some variables fell onto “Medium” category. Some recommendations are offered to help the improvement of canteen visitors satisfaction level, which will be explained below:

  • Standarize Cleanliness Facility
    • Target: Cleanliness Aspect
    • Reason Behind it: Cleanliness aspect has the lowest total average score among the others, and it variables has all variations surpassing 0.5. Especially in the cleaning_facility variable, this shows differences in students experience between canteen shifts. 
  • Product Quality and Price Control
    • Target: Product Aspect
    • Reason Behind it: Some variable is fell into “Medium” category (e.g., price_vs_quality, affordable_price), which leads to room for an improvement for an offered product. The analysis results from variables such as price_vs_quality show that the perception of value-for-money is still weak and inconsistent among respondents. 
  • Equalizing Standard at Each Campus
    • Target: All Aspect
    • Reason Behind it: Segmentation analysis has showed a significant differences in average scores (e.g., Binus Semarang have the highest variations). Meanig that service and facility standards need to be harmonized so that the student experience is more consistent.  

Steps for Implementations

Every plan need an execution steps to accomplish the changes for improvement. This action plan may useful for some period of time such as:

Short Term (1 - 3 month)

  • Audit and Manual Monitoring 
    • Apply some checklist of daily cleaning inspections for tables, chairs, and utensils.
    • Add cleaning support staff during busy hours (e.g., lunch).
  • Quick Roll Call 
    • Morning roll call and every shift change in private with cleaning staff to ensure the hygiene standard & customer service are on track.
  • Product Adjustment 
    • Putting a QR codes next to food/snack shelf to obtain reviews directly from consumers.
    • Quick review for a new product in cafeteria, for example: give a quick rate from 1 to 5 and short statements like “Food worth the price.”, “The food is fresh when it gets to you.”, etc.)

Medium Term (3 - 6 month)

  • Equalizing Standard at Every Campus 
    • Establish SOPs (Standard Operating Procedures) for applying QR codes for quick reviews on new food shelves and encourage workers at morning roll call and shift change times to remind customers to fill out reviews every time they purchase new items. With this step, it’ll be saving much more time to conducting an audit.

Long Term (6 - 12 month)

  • Infrastructure Improvement 
    • Seeing the varied responses of students regarding hygiene facilities, which indicate differences in facilities between cafeterias, it is necessary to make improvements to the infrastructure. This is important in order to provide comfort to students and ensure that they experience the same facilities regardless of the cafeteria location.
  • Continuous evaluation 
    • Seeing the varied responses of students regarding hygiene facilities, which indicate differences in facilities between cafeterias, it is necessary to make improvements to the infrastructure. This is important in order to provide comfort to students and ensure that they experience the same facilities regardless of the cafeteria location.

Closing: Expected impact

The implementation of the recommendations that have been formulated is expected to improve the consistency of hygiene standards, product quality, and uniformity of experience across campuses. With this step, student satisfaction levels can increase significantly while strengthening the image of the cafeteria as a facility that is comfortable, hygienic, and meets student needs.